Title: Diabetic Retinopathy Classification using Transfer Learning and Data Augmentation
Description: This project aims to classify diabetic retinopathy images into five classes using a Convolutional Neural Network (CNN) built with Transfer Learning Inception model. The classes are as follows: [No Diabetic Retinopathy, Mild, Moderate, Severe, and Proliferative DR]. The dataset used in this project was downloaded from Kaggle, and it contains a large number of retinal images. The images were divided into folders based on the corresponding class using the data information in csv file (the data labels). The dataset was then augmented using data augmentation techniques such as rotation, flipping, and zooming to increase the amount of training data. The CNN model was then trained on the augmented dataset using the transfer learning Inception model. Finally,the ROC curve was plotted to evaluate the model's performance.
This project demonstrates the use of transfer learning and data augmentation techniques to improve the accuracy of a CNN model. It also showcases the importance of pre-processing and organizing large datasets to efficiently train a model. This project can be useful for medical professionals and researchers interested in detecting diabetic retinopathy and developing automated diagnosis systems. The code and dataset can be accessed and used by anyone interested in the project